From 905991bf83a858a89e8d2d27abd012a76ef5bc38 Mon Sep 17 00:00:00 2001 From: Alan Paxton Date: Wed, 1 Nov 2023 16:41:49 +0000 Subject: [PATCH] Fix text per review comments, remove TODO --- docs/_posts/2023-09-12-java-jni-benchmarks.markdown | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/docs/_posts/2023-09-12-java-jni-benchmarks.markdown b/docs/_posts/2023-09-12-java-jni-benchmarks.markdown index df2e41f2c33b..3133f6a935fd 100644 --- a/docs/_posts/2023-09-12-java-jni-benchmarks.markdown +++ b/docs/_posts/2023-09-12-java-jni-benchmarks.markdown @@ -156,7 +156,7 @@ of result. the access is presumably word by word, using normal Java mechanisms. -![Copy out JNI Get - TODO - replace the plots](./jni-get-benchmarks/fig_1024_1_copyout_nopoolbig.png). +![Copy out JNI Get](./jni-get-benchmarks/fig_1024_1_copyout_nopoolbig.png). ### PutJNIBenchmark @@ -284,7 +284,7 @@ Luckily this is easy to fix. In the Java API (JNI) implementation: In the case where the `PinnableSlice` has succesfully pinned the data, this saves us the intermediate copy to the `std::string`. In the case where it hasn't, we still have the extra copy so the observed performance improvement depends on when the data can be pinned. Luckily, our benchmarking suggests that the pin is happening in a significant number of cases. -On discussion with the RocksDB core team we understand that the core `PinnableSlice` optimization is most likely to succeed when pages are loaded from the buffer cache, rather than when they are in `memtable`. There is also a suggestion that with some coding effort, it might be possible to successfully pin in the buffer cache as well. This would likely improve the results for these benchmarks. +On discussion with the RocksDB core team we understand that the core `PinnableSlice` optimization is most likely to succeed when pages are loaded from the block cache, rather than when they are in `memtable`. And it might be possible to successfully pin in the `memtable` as well, with some extra coding effort. This would likely improve the results for these benchmarks.